, S. E. de Mink

(1)

Mem. S.A.It. Vol. 84, 171

SAIt 2013 c

Memoriedella

Massive binary stars and self-enrichment of globular clusters

R. G. Izzard

¹

, S. E. de Mink

^?2,3

, O. R. Pols

⁴

, N. Langer

¹

, H. Sana

⁵

, and A. de Koter

⁵

1

Argelander Institut f¨ur Astronomy, Universit¨at Bonn, Germany.

2

Space Telescope Science Institute, Baltimore, Maryland, U.S.A.

3

Johns Hopkins University, Baltimore, Maryland, U.S.A.

4

Department of Astrophysics/IMAPP, Radboud University Nijmegen, The Netherlands.

5

Astronomical Institute Anton Pannekoek, University of Amsterdam, The Netherlands.

Abstract.

Globular clusters contain many stars with surface abundance patterns indicat- ing contributions from hydrogen burning products, as seen in the anti-correlated elemental abundances of e.g. sodium and oxygen, and magnesium and aluminium. Multiple gener- ations of stars can explain this phenomenon, with the second generation forming from a mixture of pristine gas and ejecta from the first generation. We show that massive binary stars may be a source of much of the material that makes this second generation of stars.

Mass transfer in binaries is often non-conservative and the ejected matter moves slowly enough that it can remain inside a globular cluster and remain available for subsequent star formation. Recent studies show that there are more short-period massive binaries than previously thought, hence also more stars that interact and eject nuclear-processed material.

1. Introduction

The abundance correlations and helium enrich- ment observed in globular cluster stars im- ply that proton-burning reactions are respon- sible (Prantzos et al. 2007, and many contri- butions to this volume). Hot hydrogen burning makes helium, nitrogen and aluminium, while destroying oxygen, carbon and magnesium, as required in models of self-enrichment in glob- ular clusters. However, the number of stars in a second, or further, generation is often simi- lar to or exceeds the number in the first gen- eration (Carretta et al. 2009), and the amount of nuclear-processed material currently in their atmospheres is similar to, or larger than, that present in the atmospheres of the first stellar generation. It is not clear how so much nuclear-

?

Hubble fellow.

processed mass can end up in the second gen- eration of stars. Four main channels have been investigated to date:

1. Massive Asymptotic Giant Branch

(AGB) stars are the canonically accepted

prime candidates for self-enrichment (Ventura

et al. 2001). During their thermally-pulsing

AGB (TPAGB) phase, hot-bottom burning ef-

fectively cycles the whole stellar envelope

through a hot hydrogen burning shell. A star

of mass 4 M . M . 10 M ejects about (M −

1) M of nuclear-processed material, which is

about 10% of the mass of the whole stellar

generation. This does not take into account

binary interaction which reduces the nuclear-

processed TPAGB mass yield (Izzard 2004)

while allowing for significant helium enrich-

ment (Vanbeveren et al. 2012).

(2)

172 R.G. Izzard et al.: Massive binaries and GC self-enrichment 2. Rapidly rotating massive stars also eject

hydrogen-burned material if they spin fast enough (Decressin et al. 2007). Rotational mixing transports material from the hot stel- lar core to the surface where it is ejected if the star exceeds its critical rotation rate. This is predicted to happen in some stars (de Mink et al. 2013) although the number of rapidly ro- tating stars is such that only 3% of the mass of all massive stars is ejected in this manner (de Mink et al. 2009b).

3. Stellar mergers in dense cores of globu- lar clusters may also contribute to the reservoir of nuclear processed material (Glebbeek et al.

2009) although this channel probably does not contribute enough mass to make the second generation of stars (Sills & Glebbeek 2010).

4. Massive binary stars are another source of nuclear processed material, as we explore in the following.

2. Massive binary stars

While there is some doubt about whether most stars are in multiple stellar systems, we can be sure that most stars with masses exceed- ing about 2 M live with a companion star (Kouwenhoven et al. 2007; Raghavan et al.

2010; Fuhrmann & Chini 2012). Just as im- portantly, the latest estimate of the O-type binary-period distribution in young, open clus- ters shows that more of them are close, i.e. li- able to interact by mass transfer, than previ- ously thought (Sana et al. 2012). Only about 29% of O-type stars evolve as single stars: the rest either have their envelope stripped (33%), merge (24%) or accrete mass (14%).

Because stars expand as they age, in a close binary the initially more massive (pri- mary) star overflows its Roche lobe first, trans- ferring mass onto the (initially less massive) secondary (Fig. 1). Material flows through the first Lagrange point onto the companion, car- rying with it both the chemical signature of the primary star and angular momentum. The transferred mass settles onto the surface of the secondary, spinning it up, but – at least initially – not greatly altering its chemical abundance because material near the surface of the pri-

mary is never hot enough for nuclear reactions to be efficient.

Accretion and spin up continues until the mass of the secondary increases by about 10%, at which point it rotates so fast that material at its equator is unbound (Packet 1981). Any fur- ther mass transferred by Roche-lobe overflow is ejected from the binary system at a velocity which is low compared to the proto-globular cluster ejection speed. This material may be re- tained in the cluster for further star formation.

As the primary continues to transfer mass, it loses its unburned envelope and material orig- inally deep inside the star, which has under- gone nuclear burning, is exposed at the stellar surface. First, layers burned by the CN cycle, then CNO, and later NeNa and MgAl cycles, are transferred through the Lagrange point and ejected from the binary system. Detailed bi- nary evolution models suggest that about three quarters of the transferred mass is ejected from a close binary system, i.e. an accretion effi- ciency less than about 0.25 (de Mink et al.

2009b), the binary-star physics remains highly uncertain and its study continues (e.g. van Rensbergen et al. 2011; de Mink et al. 2013).

While the binary-star scenario has not yet been explored in detail, it is observed in na- ture. The binary star RY Scuti is ejecting material rich in helium and nitrogen, and poor in oxygen and carbon, at a velocity of about 50 km s

⁻¹

(Smith, Gehrz, & Goss 2001) i.e. more slowly than a stellar wind or the escape speed of a young globular clus- ter. Further examples of binary mass trans- fer include the Algol systems (van Rensbergen et al. 2011), X-ray binaries (Flannery & Ulrich 1977) and Wolf-Rayet binaries (Petrovic et al.

2005) which must also be products of non- conservative mass transfer.

It is clear that a copious amount of material

is ejected from interacting binary stars, much

of which has been processed by nuclear burn-

ing. We estimate that as much as 13% of the

mass of a generation of stars can be ejected

in massive binaries, an amount similar to that

ejected from rapidly rotating massive stars and

AGB stars combined (de Mink et al. 2009b).

(3)

R.G. Izzard et al.: Massive binaries and GC self-enrichment 173

Hydrogen-burned ashes: ejected ...new star formation?

a

b

Fig. 1. Schematic view of Roche-lobe overflow in a massive binary system. (a) At the start of Roche-lobe overflow, the primary star (left) overflows its Roche lobe and transfers material to the secondary (right). (b) By the end of Roche-lobe overflow, the secondary has accreted unburned material while hydrogen-burned material from deep inside the primary has been ejected from the binary system and may mix with other sources of interstellar gas from which a subsequent generation of stars may form.

3. Frascati-fuelled Perspective

It is unlikely that anyone would bet more than a bottle of Frascati’s finest white wine on any single one of the proposed scenarios for globular cluster self-pollution being the only source of mass for a second generation of stars.

Massive AGB stars are generally considered the best candidate because they can process

material through hot hydrogen-burning prior to its ejection in a slow wind, although if third dredge up happens in these stars they may not be responsible (although see Yong et al. 2008).

The mass range which contributes to clusters

is unclear also, are super-AGB stars candidates

(D’Ercole et al. 2012)? Rapidly rotating mas-

sive stars certainly exist, but their total ejected

mass is not enough even assuming – realisti-

(4)

174 R.G. Izzard et al.: Massive binaries and GC self-enrichment cally? – that they are all rapid rotators (de Mink

et al. 2009b). Binary stars may eject enough mass to satisfy the requirements of a second stellar generation, but quite how conservative is binary mass loss is not clear even after many decades of study (e.g. de Mink et al. 2007, and references therein). The competition be- tween star formation and cluster gas ejection is also relevant because massive stars evolve quickly relative to AGB stars. It may be that massive-star ejecta escapes from the globular cluster before forming any new stars (see e.g.

Charbonnel et al. and other contributions to this volume).

Uncertainties in stellar physics, e.g. mass- loss rates, mixing rates and nuclear reaction rates, affect stellar yield predictions consider- ably (e.g. Ventura & D’Antona 2005; Izzard et al. 2007; Stancliffe & Jeffery 2007; de Mink et al. 2009a; Meynet et al. 2013; and many oth- ers). The magnesium-aluminium negative cor- relation is particularly difficult to reproduce be- cause it requires proton capture at temperatures which massive stars are unable to reach, while such burning is possible in massive AGB stars (Ventura et al. 2011). Still, the massive-binary channel remains relatively unexplored and a serious contributor to the mass that makes the second generation of stars in globular clusters.

Acknowledgements. RGI thanks the conference or- ganisers, the Alexander von Humboldt Foundation for supporting his work and Richard Stancliffe for a critical reading of the manuscript.

References

Carretta, E., et al. 2009, A&A, 505, 117 de Mink, S. E., et al. 2009a, A&A, 497, 243 de Mink, S. E., et al. 2013, ApJ (in press),

ArXiv 1211.3742

de Mink, S. E., Pols, O. R., & Hilditch, R. W.

2007, A&A, 467, 1181

de Mink, S. E., Pols, O. R., Langer, N., &

Izzard, R. G. 2009b, A&A, 507, L1

Decressin, T., Meynet, G., Charbonnel, C., Prantzos, N., & Ekstr¨om, S. 2007, A&A, 464, 1029

D’Ercole, A., et al. 2012, MNRAS, 423, 1521 Flannery, B. P. & Ulrich, R. K. 1977, ApJ, 212,

533 Fuhrmann, K. & Chini, R. 2012, ApJS, 203, 30 Glebbeek, E., Gaburov, E., de Mink, S. E., Pols, O. R., & Portegies Zwart, S. F. 2009, A&A, 497, 255

Izzard, R. G. 2004, Memorie della Societa Astronomica Italiana, 75, 754

Izzard, R. G., et al. 2007, A&A, 466, 641 Kouwenhoven, M. B. N., Brown, A. G. A.,

Portegies Zwart, S. F., & Kaper, L. 2007, A&A, 474, 77

Meynet, G., et al. 2013, ArXiv e-print 1301.2487

Packet, W. 1981, A&A, 102, 17

Petrovic, J., Langer, N., & van der Hucht, K. A.

2005, A&A, 435, 1013

Prantzos, N., Charbonnel, C., & Iliadis, C.

2007, A&A, 470, 179

Raghavan, D., et al. 2010, ApJS, 190, 1 Sana, H., et al. 2012, Science, 337, 444 Sills, A. & Glebbeek, E. 2010, MNRAS, 407,